Advanced object classification using image texture descriptors
An innovative approach to object classification using a combination of deep learning, wavelength selection, and texture descriptor vectors to enhance accuracy in image analysis.
Overview
This solution presents an advanced method for object classification in images by integrating multiple characterization techniques. By combining deep learning networks, wavelength selection, texture descriptor vectors, and contextual image information, this approach enhances the accuracy and effectiveness of object identification. The method is particularly beneficial for applications requiring precise feature detection, such as in remote sensing or quality control.
Technical specifications
- Characterization Methods: Utilizes co-occurrence matrices and data transformations for neural networks.
- Wavelength Analysis: Focuses on precise pixel values crucial for accurate wavelength-based characterization.
- Texture Descriptors: Applies various transformation space methods for enhanced pattern recognition.
- Data Enrichment: Incorporates sensor data and other assessments to highlight subtle patterns.
- Model Training: Employs deep learning and machine learning methods, using data transformations and feature selection to optimize classification models.
Technology readiness level
The technology is currently at TRL 6, indicating a prototype demonstrated in a relevant environment. Further validation is planned to ensure robust ground truth data under real-world conditions.
About Universidad Antonio Nariño
Universidad Antonio Nariño is a private Colombian higher-education institution with a comprehensive, nationally distributed profile and more than 90 undergraduate programs across 26 cities. Its multi-campus structure gives corporate R&D teams access to regional talent, facilities, and local partners rather than a single-site research base. Specialized laboratories, including clinical simulation infrastructure, support experimental work, technical services, and external use, while a dedicated technology-transfer function manages patents, prototypes, industrial design, technology acceleration, and spin-off pathways. Research draws on institutional support and competitive external funding, including MinCiencias and Colombia’s General System of Royalties. The result is a practical route for companies seeking collaboration, testing, advisory services, or knowledge transfer.
delivered to your inbox.